[Assessment for spatial uncertainty of daily minimum temperature by using sequential Gaussian simulation].

Understanding daily minimum temperature is of great importance for assessing low temperature damages to crops and guiding people to take timely remedial measures to ensure food security. Kriging is a widely used technology for mapping the spatial distribution of the near-surface temperature. However, the smoothing effect, commonly found in the Kriging maps, leads to low values to be overestimated and high values to he underestimated. For daily minimum temperature on Hainan Island which was affected by cold air on December 12, 2011, cross-validation was adopted to evaluate the prediction accuracy of ordinary Kriging (OK) and Kriging with external drift (KED). The spatial distribution maps of daily minimum temperature on Hainan Island on December 12, 2011 produced by OK and sequential Gaussian simulation (SGS) were compared. Results showed that the prediction accuracy of KED (r = 0.86) was not superior to OK (r = 0.86) significantly. SGS could generate multiple equiprobable simulation realizations, and the distribution and variance function of the original data could be reproduced in the realizations. The simulation realizations generated by SGS overcame the smoothing effect of Kriging and could more truly reflect the spatial distribution of minimum temperature on the day on Hainan Island. In the region where daily minimum temperature was low, and the temperature change was small, the conditional variance of the SGS results was less than the ordinary Kriging variance. Spatial uncertainty of a potential chilling damage area could be quantified by multiple simulation realizations generated by SGS. SGS was a valuable tool for assessing agro-meteorological disasters caused by low temperature.